-
Notifications
You must be signed in to change notification settings - Fork 3.8k
Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
80494: server: add server-wide limit on addsstable send concurrency r=dt a=dt While we generally prefer recipient-side limiting to better utilize all available nodes' capacity, this adds an extra layer of limiter just in case we find ourselves needing it. It defaults to unlimited. Release note (sql change): The cluster setting bulkio.ingest.sender_concurrency_limit can be used to adjust the concurrency at which any one SQL node, across all operations that it is running such as RESTORES, IMPORTs, and schema changes, will send bulk ingest requests to the KV storage layer. Epic CRDB-2264 81177: ccl/sqlproxyccl: invoke rebalancing logic during RUNNING pod events r=JeffSwenson a=jaylim-crl #### ccl/sqlproxyccl: invoke rebalancing logic during RUNNING pod events This commit invokes the rebalancing logic during RUNNING pod events as part of the pod watcher. Since the rebalancing logic depends on the tenant directory, the pod watcher will now only emit events once the directory has been updated. This is done for better responsiveness, i.e. the moment a new SQL pod gets added, we would like to rebalance all connections to the tenant. Note that the Watch endpoint on the tenant directory server currently emits events in multiple cases: changes to load, and changes to pod (added/modified/ deleted). The plan is to update the tenant directory server to only emit events for pod updates. The next commit will rate limit the number of times the rebalancing logic for a given tenant can be called. At the same time, we introduce a new test static directory server which does not automatically spin up tenants for us (i.e. SQL pods for tenants can now be managed manually, giving more control to tests). #### ccl/sqlproxyccl: rate limit the number of rebalances per tenant This commit rate limits the number of rebalances per tenant to once every 15 seconds (i.e. 1/2 of the rebalance loop interval). The main purpose of this is to prevent a burst of pod events for the same tenant causing multiple rebalances, which may move a lot of connections around. Release note: None 81582: sql: add is_grantable column to SHOW GRANTS FOR role r=richardjcai a=ecwall refs #73394 Release note (sql change): Add is_grantable column to SHOW GRANTS FOR role to be consistent with other SHOW GRANTS commands. 81776: bazel: bump size of `rangefeed` test r=rail a=rickystewart This has timed out in CI. Release note: None Co-authored-by: David Taylor <[email protected]> Co-authored-by: Jay <[email protected]> Co-authored-by: Evan Wall <[email protected]> Co-authored-by: Ricky Stewart <[email protected]>
- Loading branch information
Showing
36 changed files
with
2,828 additions
and
1,979 deletions.
There are no files selected for viewing
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Oops, something went wrong.